The incidence of multidrug-resistant (resistant to isoniazid and rifampin) and extensively drug-resistant (XDR, additional resistance to fluoroquinolones [FQ] and second-line injectables [SLI]) tuberculosis (TB) is increasing, causing serious illness with high mortality. Rapid diagnosis of drug-resistant TB is critical to improving patient outcomes and preventing further acquisition of resistance to other drugs. Rapid molecular tests have revolutionized the diagnosis of drug-resistant TB by shortening the time to diagnosis from 3-6 weeks to <1 day. Commercial tests have now been approved by the World Health Organization to diagnose multidrug resistant TB (MDR TB). However, these methods fail to identify a substantial proportion of Mycobacterium tuberculosis isolates shown to be resistant to FQ or SLI by culture-based phenotypic tests (the gold standard) for at least two reasons: 1) current molecular assays have inadequate resolution to identify small sub-populations of resistant bacilli harboring mutations mixed with susceptible sub-populations (hetero-resistance), which emerging data suggests is common at least for FQ and 2) there are additional genetic markers of resistance that have yet to be identified. The overall goal of this proposal is to fundamentally advance molecular diagnostics for pre-XDR and XDR TB. We will pursue parallel research aims that represent a complementary and systematic approach to improving the diagnosis and management of drug-resistant TB. In Aim 1, we will perform targeted next-generation deep sequencing of all known gene regions associated with Mycobacterium tuberculosis drug resistance using stored culture isolates and clinical samples from patients with MDR, pre-XDR and XDR TB being evaluated at two leading MDR TB treatment centers in Manila, Philippines. Results will be used to characterize the frequency of hetero-resistance for different anti-TB drugs and the ability of current molecular assays to identify resistance in the presence of mixed bacillary populations. We will also assess the impact of hetero-resistance on treatment outcomes. In Aim 2, we will perform whole genome next-generation sequencing in cases (FQ- and SLI-resistant isolates without a known mutation) and controls (FQ- and SLI-susceptible isolates) to identify novel genetic markers of FQ and SLI resistance. We will develop prediction models to quantify improvements in diagnostic accuracy for FQ and SLI resistance when adding the novel genetic markers and perform experiments to identify which novel mutations directly cause FQ or SLI resistance. When the proposed analyses are complete, we will have produced a comprehensive description of reasons for discordance between phenotypic and genotypic tests for resistance to FQ and SLI and evidence on the impact of hetero-resistance on treatment outcomes. In addition, we will have generated a list of novel mutations that improve prediction of and directly cause FQ and SLI resistance. The findings will contribute to improved diagnostics and management strategies for drug-resistant TB.